fix(docs): correct misleading implant cleanup statement across third-party executors

[Copilot]

The docs incorrectly claimed implant directories are "deleted on assets' restart" — a client confirmed runtimes/ and payloads/ persist indefinitely after restart, contradicting this.

Changes

docs/deployment/ecosystem/executors.md

Replaced the inaccurate single-sentence cleanup claim in all four third-party executor sections (Tanium, CrowdStrike, SentinelOne, Palo Alto Cortex) with a structured bullet list:

• Each implant directory is unique per inject.
• Old implants are periodically cleaned up by the OpenAEV platform based on the clean-implant-interval configuration parameter (default: 8 hours).
• This cleanup is performed server-side — implant directories are not automatically deleted on endpoint restart.

The clean-implant-interval parameter already existed in each executor's config table but was never referenced in the prose.

docs/usage/openaev-agent.md

Expanded the bare "Execution cleanup and directory pruning" feature bullet into sub-bullets covering:

• Agent-side pruning of runtimes/implant-XXXXX on implant process exit
• payloads/ directory cleanup after payload delivery
• Edge case: residual directories when the agent is offline or a process is killed externally, with the server-side clean-implant-interval sweep as the fallback